[lmi-commits] [lmi] master 04cc85e 1/2: Try to provoke division by zero in round_to ctor-initializer

[Greg Chicares]

branch: master
commit 04cc85ef1b2b217ccdd5cc3997ec79e018f0fbbc
Author: Gregory W. Chicares <address@hidden>
Commit: Gregory W. Chicares <address@hidden>

    Try to provoke division by zero in round_to ctor-initializer
    
    Added a test to guard this ctor against the problem discussed in the
    last several commits for class paginator. It seems unlikely that 10^N
    will equal zero for any finite negative integer N.
---
 round_to_test.cpp | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/round_to_test.cpp b/round_to_test.cpp
index 4b5b62a..a14c631 100644
--- a/round_to_test.cpp
+++ b/round_to_test.cpp
@@ -28,6 +28,7 @@
 #include "test_tools.hpp"
 
 #include <algorithm>                    // max()
+#include <climits>                      // INT_MIN
 #include <ios>
 #include <iostream>
 #include <ostream>
@@ -543,6 +544,16 @@ int test_main(int, char*[])
     BOOST_TEST(2 == round1.decimals());
     BOOST_TEST(r_to_nearest == round1.style());
 
+    // Try to provoke division by zero in ctor-initializer.
+    //
+    // nonstd::power() negates a negative exponent, but negating
+    // INT_MIN constitutes UB, so use 1 + INT_MIN.
+    BOOST_TEST_THROW
+        (round_to<double>(1 + INT_MIN, r_to_nearest)
+        ,std::domain_error
+        ,"Invalid number of decimals."
+        );
+
     // The software default rounding style and the hardware rounding
     // mode may be either synchronized or not, so test both ways.
     std::cout << "  Default style synchronized to hardware mode:\n";